Big Bear AI (BigBear.ai): BBAI stock and national security AI

Revenue up 31%. A $91.9 million loss. A share price down about 62% in a year. BigBear.ai managed all three in the same quarter, and I think that odd combination tells you more about Big Bear AI than any of its press releases do.

The company sells AI-powered decision intelligence tools to Washington: software for the Army, for border officers, for planners who move ships and spare parts around. Its ticker is BBAI. People keep comparing it with Palantir, which is flattering in one sense and quite unfair in another.

One housekeeping note. Every figure here is as of November 2, 2026, unless stated otherwise, and nothing in this piece is investment advice.

What BigBear.ai actually sells to national security buyers

Think of it as a toolbox, not a product. A handful of very large government buyers pick the tools they need, and in 2026 one of those tools started doing most of the heavy lifting.

Ask Sage: secure generative AI for defense

Ask Sage is generative AI for rooms where ChatGPT is banned. Defense units, intelligence teams, tightly regulated agencies: they get AI models that work on sensitive material without anything leaking to a public service.

BigBear.ai paid $250 million in cash for it. The deal closed on December 31, 2025 — the biggest purchase the company has ever made. You can see it in the margins almost at once. Software earns more per dollar than consulting hours, so gross margin (what is left after direct costs) went from roughly 22% in 2025 to 31-33% across the first three quarters of 2026.

Border, airports and digital identity

Leg two is identity and borders. Pangiam joined in February 2024. Its co-founder, Kevin McAleenan, used to run US Customs and Border Protection; since January 2025 he has run BigBear.ai.

Trueface lives here. Airports and ports of entry use it for real-time face matching, a big part of modern airport security. Management, as quoted by The Motley Fool, claims 99.1% accuracy against 12 million identities. A vendor number, not an audit. Still, you can see why customs officers like it.

Digital twins, supply chain and decision intelligence

The third leg is older and plainer. ProModel builds digital twins: computer copies of a factory, a port or an Army unit, so planners can test a decision before making it. ConductorOS pushes AI deployment out to the edge, onto mission-critical devices sitting somewhere with a weak signal. Much of this is machine learning, predictive models and advanced analytics for logistics. Which spare part runs out first? Which ship arrives late?

VANE reads foreign news for hints of what adversaries are planning; the Pentagon's Chief Digital and Artificial Intelligence Office gave it a prototype contract in February 2025. There is cybersecurity work as well, such as AI-powered checks of software for hidden flaws.

Who pays? The US Army, the Department of Defense, the intelligence community, Homeland Security and its border agency, the Coast Guard, NASA and a few more.

From SPAC to Ask Sage: how BigBear.ai Holdings got here

The family tree goes back to 1988. Private equity firm AE Industrial Partners stitched several government contractors together, and in December 2021 the result went public through GigCapital4, a SPAC: a listed shell company whose only job is to buy a private one.

The playbook since then has been to buy skills rather than build them. ProModel in 2022. Pangiam in 2024. Ask Sage at the very end of 2025, then CargoSeer, which inspects cargo with AI, in January 2026. Headquarters are in McLean, Virginia, and the annual report counted 579 employees at the end of 2025. Kevin McAleenan took over from Mandy Long as CEO in January 2025.

BBAI earnings call: what the Q3 2026 numbers really say

Growth was the headline of the October 27 earnings call. Fair enough: it is real. The rest of the income statement is more sobering.

Sources: BigBear.ai earnings releases and FY2025 10-K. Growth rates are year over year.

Revenue rose 31%, to $43.6 million from $33.1 million, according to BigBear.ai's third-quarter earnings release filed with the SEC that day. Analysts had pencilled in about $38.9 million, per Yahoo Finance. Management lifted the 2026 forecast to $150-165 million; the old range started at $135 million.

Now the loss. That works out to $0.19 per share, and most of the $91.9 million was a $76.7 million goodwill impairment. Goodwill is what a buyer pays above the book value of what it buys. Impairing it means admitting part of that premium has evaporated. No cash left the building — but it is an accounting confession about past deals. Strip it out and adjusted EBITDA, a rough gauge of operating cash profit, was still minus $9.1 million.

Backlog needs translating too. About $330 million sounds big next to a $43.6 million quarter. Only some 23% of it is funded, though, meaning Congress has actually appropriated the money. The rest is contract ceilings and options. Some will become revenue; some will not.

The nine-month figures make the point bluntly. January to September 2026 brought in $114.8 million, against $100.4 million a year earlier. The net loss over the same stretch: $174.4 million. Roughly a dollar and a half lost for every dollar of sales.

And context matters. Revenue fell 38% in the fourth quarter of 2025, when Army programs slowed, and 2025 closed with a $293.9 million net loss. So this year's growth is a rebound from a weak base, helped a lot by one big acquisition. Worth knowing before paying for it.

Why BBAI stock price fell 62% in a year

The last close before this was written: $2.60, on October 30. That values the whole company at roughly $1.25 billion. Over 52 weeks the shares lost 62.4%; since January, about 52%, per Yahoo Finance data on November 2. Twelve months ago they touched $7.40.

Growth alone does not explain a fall this steep. Three worries do.

Dilution, first. At the end of 2025 there were about 437 million shares. By September 30, 2026, there were 480 million — 9.9% more in nine months, mostly because convertible notes turned into stock. Anyone who held through that saw their slice of the company shrink.

Then the losses, quarter after quarter. And then scale. A business expecting $150-165 million of revenue this year is small fry in defense technology, where single programs run into the billions.

Short sellers have taken the other side in size: about 28.6% of the tradable shares are sold short. Simply Wall St puts the stock's beta near 3.3, so on a bad day for the market it usually has a much worse one.

Wall Street is lukewarm. Cantor Fitzgerald stayed at Neutral on October 28 and trimmed its target from $4 to $3. Citi said Market Perform the same day. H.C. Wainwright is still at Buy. It did cut its target from $8 to $6 in March. Weiss Ratings? Sell. My reading of that mix: analysts want proof, and they have not decided yet.

BigBear.ai vs Palantir: same AI label, different businesses

Investors often treat BigBear.ai as a cheaper way into the Palantir story. The numbers say they are different kinds of company.

Palantir figures are from its Q2 2026 results; its Q3 report was due after the close on November 2. Market data as of November 2, 2026.

Palantir's revenue in a single quarter, $1.935 billion according to its second-quarter release of August 2026, is about 44 times BigBear.ai's latest quarter. Its market value is roughly 350 times larger. Palantir is also very profitable, with a GAAP net margin of 55% in that quarter, while BigBear.ai still loses money.

The clearest signal sits in the price-to-sales column. Investors pay about 72 dollars for every dollar of Palantir's annual revenue and about 9 dollars for BigBear.ai's. That gap is the market's view of the difference between a platform the Pentagon builds on and a specialist contractor that sells tools into specific programs.

The customer mix explains a lot of it. Palantir's US government revenue reached $809 million in the second quarter, up 90%, and its US commercial revenue was $764 million, up 149%. In other words, almost half of its business now comes from companies, not agencies, which gives it a second engine. BigBear.ai has no comparable commercial base; its growth still depends on winning and expanding government programs one at a time.

Margins tell the same story from another angle. Palantir reported an adjusted operating margin of 62% in the second quarter, because one platform is sold many times over. BigBear.ai's gross margin, before any operating costs, is about 31%. Each new contract brings some services work with it, and services do not scale like software.

There is a fair counterpoint. A low multiple also means low expectations. If BigBear.ai keeps growing at 20-30% and narrows its losses, a small base can move the percentages quickly, and the stock does not need Palantir-style growth to justify today's price. They answer different questions for an investor, and one cannot stand in for the other.

Risks: homeland security budgets, dilution and a restatement

Start with concentration, because it already bit once. The FY2025 annual report shows that customers each worth more than 10% of revenue added up to 51% of sales. When Army volume dropped in late 2025, quarterly revenue fell 38%. One program, one quarter, one hole in the income statement.

Problem two is Washington's calendar. Almost every dollar comes from defense and homeland security agencies. A late appropriation, a shutdown, new priorities in Congress — any of them can freeze work that looked signed and sealed. The timing of defense contracts matters as much as winning them.

Third, the accounts. In March 2025 BigBear.ai told investors not to rely on its earlier financial statements. The reason: it had accounted for its 2026 convertible notes the wrong way. It restated them and flagged a material weakness in internal controls. A shareholder lawsuit followed, filed in federal court in Virginia. Where that case stands today I could not confirm; the latest 10-Q is the place to check.

And the familiar one: while losses continue, more dilution stays possible.

What would have to go right for Big Bear AI in 2027

A credible bull case exists — it just depends on execution.

Since the second quarter, the company says, it has signed more than 25 new contracts worth over $225 million in total contract value. Border agencies, the FBI, the Coast Guard and NASA are on the list. None of that is revenue yet. It has to become funded backlog first, then billed work.

Ask Sage has to keep spreading through defense and intelligence teams, too. It is the one piece with software economics.

Money, at least, is not the constraint right now. Cash and investments stood near $399 million at the end of September. Long-term debt: zero. About $17 million of convertible notes come due in December 2026, and management expects to walk into 2027 debt-free.

If you track BBAI, the press releases matter less than four numbers. Funded backlog. Gross margin above 30%. Adjusted EBITDA heading toward zero. A share count that stops climbing. Wins in border control and critical infrastructure would show up in all four.

The verdict on BigBear.ai in late 2026

Big Bear AI has paying government customers and a turnaround that shows up in the 2026 numbers. It is also small, unprofitable and dependent on a handful of programs, and its recent growth was largely bought through the Ask Sage acquisition.

At about $2.60 a share, the market is pricing that doubt, not the AI label. Whether that is too pessimistic depends on what happens to funded backlog and the share count over the next few quarters, not on how often the company is compared with Palantir.

So the useful question is whether BigBear.ai can grow into a profitable specialist, whatever anyone calls it. Watch the next earnings call with that question in mind.

Marco Lucchetti

Marco Lucchetti is a senior content strategist and blockchain analyst at Plisio. With over 7 years of experience in cryptocurrency research, DeFi protocols, and payment technologies, Marco specializes in creating clear, data-driven content for a global crypto audience. His work focuses on transaction tracing, crypto compliance, and the future of blockchain infrastructure.